›Released: July 19, 2026 — source: marktechpost.com/perplexity-ai-releases-wandr-an-open-benchmark-evaluating-research-agents-that-must-search-wide-and-deep
›WANDR is an open benchmark and evaluation harness for research agents.
›Built around 500 realistic, challenging data-collection tasks for knowledge work.
›Aims to test agents on wide and deep collection patterns at professional scale, such as competitive mapping, due diligence, and literature review.
›Complements Perplexity's DRACO benchmark for deep research, which focuses on accurate, complete, objective long-form reports.
›Perplexity achieved 0.447 soft F1 at the xhigh setting, with costs ranging from $0.03 to $324.83 per task.
›Highlights that discovery is a structural bottleneck and turning a usable page into complete evidence is challenging.